Green Plastics Patent Abstracts with Binary Labels

Patents_Green_Plastics contains 11,196 patent abstracts labeled for green plastics classification. The dataset was created by cwinkler in 2023, filtering abstracts from the BIGPATENT source based on polymer-related terms. It features a binary label indicating whether an abstract pertains to 'Green Plastics' (1) or 'Not Green Plastics' (0).

Use Cases

• Train a text classifier using the title and label fields to predict if a patent abstract describes green plastics.
• Analyze the linguistic patterns in the title field that correlate with the green plastics label.
• Fine-tune a language model for binary classification on 11,196 labeled patent abstracts.
• Investigate the distribution and characteristics of abstracts labeled as 1 versus 0 for sustainability research.

Strengths

• Provides 11,196 labeled text samples for a specific sustainability domain.
• Clear binary classification task with labels 0 and 1 for 'Not Green Plastics' and 'Green Plastics'.
• Derived from the established BIGPATENT dataset, indicating a structured source.

Limitations

• Dataset size of 11,196 rows is moderate for deep learning applications.
• Limited to abstracts containing the terms 'plastics' or 'polymer', which may introduce selection bias.
• Lacks detailed column information beyond title and label, limiting feature analysis.

Provenance

Source

BIGPATENT dataset

Collection Method

Abstracts were filtered for 'plastics' and 'polymer' terms, then labeled for green plastics content.

Freshness

Last updated on 2023-01-16.